Skip to content

deps(deps): update zarr[cast-value-rs] requirement from <3.5.0,>=3.2.0 to >=3.4.2,<3.5.0 - #342

Open
dependabot[bot] wants to merge 1 commit into
mainfrom
dependabot/uv/zarr-cast-value-rs--gte-3.4.2-and-lt-3.5.0
Open

dependabot[bot] wants to merge 1 commit into
mainfrom
dependabot/uv/zarr-cast-value-rs--gte-3.4.2-and-lt-3.5.0

Conversation

@dependabot

@dependabot dependabot Bot commented on behalf of github Nov 4, 2026 •

Copy link
Copy Markdown
Contributor

Updates the requirements on zarr[cast-value-rs] to permit the latest version.

Release notes

Sourced from zarr[cast-value-rs]'s releases.

v3.4.2

What's Changed

Dependency updates

What's Changed

New Contributors

Full Changelog: zarr-developers/zarr-python@v3.4.1...v3.4.2

Changelog

Sourced from zarr[cast-value-rs]'s changelog.

3.4.2 (2026-10-27)

Features

  • Reading data encoded with the pickle codec now requires an explicit opt-in, because unpickling data from an untrusted store can execute arbitrary code (GHSA-jqc8-qjfv-5m6r). Enable it with zarr.config.set({"array.allow_unsafe_pickle_decode": True}) or the environment variable ZARR_ARRAY__ALLOW_UNSAFE_PICKLE_DECODE=True. Without the opt-in, reading such data raises zarr.errors.PickleCodecDisabledError. numcodecs has its own opt-in (numcodecs.pickles.ALLOW_UNSAFE_PICKLE_DECODE), which must also be enabled. Writing with the pickle codec is unchanged. (#4465)

Bugfixes

  • Resizing an array with delete_outside_chunks=True now generates only the out-of-bounds chunk coordinates and deletes them with a bounded pool of workers, instead of materializing the entire old and new grids and all deletion tasks. This bounds auxiliary memory during shrinking and mixed growing/shrinking resizes while preserving retained chunks. Growing, or shrinking within the last chunk of each axis, no longer schedules any deletion work. When a deletion fails, the remaining deletions are now canceled before the error is raised; previously they ran to completion. (#4344)
  • zarr.open_array and zarr.open (sync and async) now apply the config argument when opening an existing array. Previously config was only used when a new array was created, so zarr.open_array(store, config={"read_missing_chunks": False}) silently ran an existing array with the global defaults. config is now a named parameter of both functions and of Array.open/AsyncArray.open. zarr.open raises a TypeError naming the problem when config is given for a group. (#4408)
  • Fixed several ways ZipStore could lose data or fail. A store created with mode="w" or mode="x" now keeps its entries when it is used again after close(), move(), or unpickling, instead of truncating the archive. Every method now opens the archive on first use, under a lock, so get, get_partial_values, set_if_not_exists, and clear no longer raise AttributeError on a store that was never opened, and threads sharing a store no longer open the archive twice. An unpickled store now opens its archive on first use rather than when it is unpickled, so writing a dask array to a ZipStore no longer produces an unreadable file. Where reopening an archive would silently drop its entries, the store now raises instead: RuntimeError after a failed close(), zipfile.BadZipFile if the file is not a complete zip archive, and io.UnsupportedOperation for a file object that is not readable and seekable. The type of the mode parameter now includes "x". Builds on #3593 by @​oelhammouchi. (#4450)
  • Reads through FusedCodecPipeline no longer hang in a process forked after an earlier read had started the codec thread pool. The pool is now reset in the child process after a fork. Previously the child inherited an executor whose threads did not exist, so small reads were queued on a pool nothing would run. (#4483)
  • Structured data types now default string (S, U) and raw-bytes (V) fields to zero bytes. Before, an array with a raw-bytes field failed to create without an explicit fill value, and string fields defaulted to "0". (#4495)
  • The minimum supported numcodecs version is now 0.16.6, and 0.17.0 is excluded. Those versions leave Pickle.decode enabled by default, which allows arbitrary code execution when reading an untrusted Zarr v2 array (GHSA-jqc8-qjfv-5m6r). (#4465)
  • The zarr.storage module can now be serialized with pickle and cloudpickle. Previously, code that referenced the module (for example a Dask task that used zarr.storage.ObjectStore) failed with TypeError: cannot pickle 'VerboseModule' object. (#4557)

Improved Documentation

  • Document the removal of LRUStoreCache in the v3 migration guide, and point to the experimental zarr.experimental.cache_store.CacheStore as its replacement. (#3850)
  • Added a deprecation policy to the contributing guide. It defines when a deprecation cycle is required for public API changes, a shared framework for deciding whether a removal is worthwhile, the mechanics a deprecation must follow (warning class, message contents, documentation, and changelog entry), and a minimum deprecation period of at least 6 months and at least one minor release before removal. (#4051)
  • Clarify how Zarr format 3 codec pipeline roles map to filters, serializer, and compressors when creating arrays. (#4166)
  • Correct the docstrings of zarr.save, zarr.save_array and zarr.save_group to say the functions write to any store. Previously the docstrings stated the functions wrote to only the local file system, which was inaccurate. (#4520)

Misc

3.4.1 (2026-10-08)

Bugfixes

  • Chunk sizes must now be at least 1, while an array extent may still be 0. On a zero-length axis, chunks=-1, chunks=False, chunks="auto", shards=-1 and shards=False all give chunk size 1, and a shard spanning an axis is rounded up to a multiple of the inner chunk size, so shards=-1 works when the axis length is not a multiple of it. A rectilinear chunk grid can be declared on a zero-length dimension; its chunk sizes are kept for later growth.

    FixedDimension(size=0, ...) raises ValueError. ArrayV2Metadata(chunks=(0,)) is still accepted and written as given, and is read back the way a stored 0 is read (below). create_hierarchy now builds every node before it deletes or stores anything, so a node that cannot be built leaves the store untouched. Regular chunk grid metadata reads a bool chunk size as the integer it equals and rejects a string or mapping as a chunk shape. RectilinearChunkGridMetadata (experimental) does the same for bool edges and rejects NumPy integers and floats such as 4.0 with TypeError; it used to store them as given, which produced JSON floats the specification does not allow.

    Stored metadata with a regular chunk size of 0 or JSON false (Zarr format 2 chunks, Zarr format 3 regular chunk_shape) is now read as chunk size 1, or the inner chunk size for sharded arrays, which previously could not be opened. The result does not depend on how far the axis has grown since. zarr-python wrote such sizes for arrays created with a zero-length axis until 3.4, and from 2.18.7 to 3.2.1 for an explicit chunks=(0,) or False on any axis; those arrays hold no data. On a zero-length axis the array opens silently, and appending stores chunks of size 1. On a longer axis it opens with a ZarrUserWarning that says the array holds only its fill value, how to recreate it with the wanted chunk shape (zarr.from_array(array.store, name=array.path, data=array, chunks=..., overwrite=True, write_data=False) keeps the data type, fill value, attributes and codecs), and how to keep it by storing valid metadata instead (array.update_attributes({}), then zarr.consolidate_metadata if the metadata is consolidated). JSON true, which zarr-python 3.0 and 3.2 wrote for chunks=True, is read as 1 without a warning, in regular and rectilinear chunk grids and in the inner chunk shape of any sharding codec. Rectilinear edges stored as integral JSON floats ([[4.0, 2]]), as 3.2 wrote them, are read as integers without a warning. A float anywhere else is rejected.

    Writing data to an array whose stored chunk size is 0 or false first re-reads the stored metadata and stores its repair, unless another writer has stored valid metadata since, so that other readers find the chunks written; in a ZipStore this adds a second entry for the metadata document, as every metadata update does. If the stored metadata now lays out chunks differently, the write raises ValueError asking to reopen the array and stores nothing. update_attributes and resize store the repair as before. No other operation does: a group's consolidated metadata keeps the array's metadata as it was stored until the array itself writes the repair. (#4334)

  • Nested structured data types now round-trip through Zarr format 2 metadata. Converting a NumPy dtype with field titles or subarray fields raises ValueError naming the unsupported field, instead of producing duplicate or raw-byte fields; this is a limitation of the current zarr-python dtype implementation, not of the format. Padded layouts are still converted to packed layouts, now with a ZarrUserWarning that field values are preserved but offsets and itemsize may change. These checks apply to nested fields too. New zdtypes and structured_dtypes hypothesis strategies back property tests for dtype serialization and layout conversion. (#4340)

  • DateTime64 and TimeDelta64 now normalize μs to us and keep generic-unit scale factors when converting NumPy dtypes and writing V2 and V3 metadata. V2 metadata spells a scaled generic unit explicitly, such as [2generic], because NumPy's dtype.str drops the scale. Generic temporal arrays keep their dtype parameters when CPU buffers are allocated and convert byte order correctly, and generic datetime fill values can be read back, including in structured fields. The default NaT fill value carries the scale factor, so missing chunks of an array such as m8[2us] read back as NaT on NumPy < 2.2 instead of an arbitrary count. (#4342)

  • LocalStore no longer blocks the event loop on filesystem calls. open, clear, delete, delete_dir, list, list_prefix, list_dir, move and getsize now run their disk I/O in a worker thread, as get and set already did. The LocalStore tests fail if any async store method touches the filesystem from the event loop thread. (#4353)

  • Coordinate selections, including vindex masks, and orthogonal integer-array selections no longer allocate memory proportional to the total number of chunks, so selecting a few points from an array with billions of chunks no longer raises MemoryError. The indexers keep only the occupied chunk IDs and run-end offsets (chunk_run_ends); unsorted selections may still need a sort. The dense chunk_nitems and chunk_nitems_cumsum attributes of CoordinateIndexer and IntArrayDimIndexer are still available: each is computed on first access and cached, so its identity and in-place mutations persist, and dataclasses.asdict() still materializes them. repr omits them. Accessing one still allocates memory proportional to the full chunk grid. The new zarr.core.common.ceildiv_int computes chunk counts and slice bounds in exact integer arithmetic; ceildiv keeps its floating-point behaviour. (#4356)

  • Runtime code no longer uses assert, and ruff's S101 rule keeps it out. Asserts are stripped under python -O, and two were load-bearing: GroupMetadata.from_dict asserted on node_type, which the array-to-group fallback in zarr.open relied on, and make_store asserted on mode. Both now raise proper errors, NodeTypeValidationError and ValueError. (#4363)

  • Arrays whose stored regular chunk grid mixes integers with lists of edge lengths, such as "chunk_shape": [2, [5, 10, 5]] (written by zarr 3.2.0 and 3.2.1 for chunks=(2, (5, 10, 5)), with float edges for float input), can be read again without enabling array.rectilinear_chunks. The grid is read as the rectilinear grid it describes, with a ZarrUserWarning. Re-saving the array's metadata, or writing chunks to it, stores that rectilinear grid and so requires the flag. A group's consolidated metadata keeps the array's metadata as stored.

    The array.rectilinear_chunks flag now gates reading and storing array metadata that declares a rectilinear chunk grid, including in consolidated metadata, rather than the construction of RectilinearChunkGridMetadata. An operation that would store such metadata with the flag off (creating an array, Array.resize, deleting a member of a group with consolidated metadata, create_hierarchy) raises before it deletes or writes anything, and the error names the array. (#4375)

  • Chunk specifications are now judged by the chunk normalizer alone; the separate duck-typed check for rectilinear input is gone.

... (truncated)

Commits
  • a2f2415 docs: 3.4.2 release notes (#4565)
  • d588831 chore(deps): bump setuptools from 83.0.0 to 84.1.0 (#4566)
  • 1ee442b chore(deps): bump graphql-core from 3.2.8 to 3.2.12 (#4567)
  • e28e7dc ci(big-endian): install tzdata so pip check passes on the s390x image (#4570)
  • 04dd0fb docs: create the explicit serializer example array with overwrite=True (#4569)
  • 9b7b93b feat: require opt-in to read data encoded with the pickle codec (#4465)
  • e376ace fix(api): apply config when open_array opens an existing array (#4408)
  • 332caef docs: clarify Zarr v3 codec pipeline roles (#4166)
  • a429d16 fix(storage): open ZipStore archive on first use from every method and reopen...
  • 44e89e7 chore(zarr-metadata): build 0.6.0 changelog (#4563)
  • Additional commits viewable in compare view

@dependabot dependabot Bot added dependencies Pull requests that update a dependency file python:uv Pull requests that update python:uv code labels Nov 4, 2026
@dependabot @github

dependabot Bot commented on behalf of github Nov 4, 2026

Copy link
Copy Markdown
Contributor Author

Assignees

The following users could not be added as assignees: developmentseed/data-model-developers. Either the username does not exist or it does not have the correct permissions to be added as an assignee.

Please fix the above issues or remove invalid values from dependabot.yml.

Updates the requirements on [zarr[cast-value-rs]](https://github.com/zarr-developers/zarr-python) to permit the latest version.
- [Release notes](https://github.com/zarr-developers/zarr-python/releases)
- [Changelog](https://github.com/zarr-developers/zarr-python/blob/main/docs/release-notes.md)
- [Commits](zarr-developers/zarr-python@v3.2.0...v3.4.2)

---
updated-dependencies:
- dependency-name: zarr[cast-value-rs]
  dependency-version: 3.4.2
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
@dependabot
dependabot Bot force-pushed the dependabot/uv/zarr-cast-value-rs--gte-3.4.2-and-lt-3.5.0 branch from 9c15525 to 5d5c0ff Compare November 4, 2026 15:43
@d-v-b d-v-b self-assigned this Nov 13, 2026
@d-v-b

d-v-b commented Nov 13, 2026

Copy link
Copy Markdown
Contributor

this is waiting on zarr-developers/zarr-python#4526

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

dependencies Pull requests that update a dependency file python:uv Pull requests that update python:uv code

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant